Saving frames/Image from PGIE Detector after Classification from SGIE

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) Jetson Orin Nano
• DeepStream Version 6.3.0
• JetPack Version (valid for Jetson only) 5.1.2
• TensorRT Version 8.5.22
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs) questions
• Language Python 3.8

I am working on a face recognition deepstream pipeline in python. My pipeline basic structure is as follows:
Stream->PGIE(Detector)->Tracker->SGIE(Face Embedding Model)->…

My PGIE is a face detector which crops the face detected and pass it as input to SGIE (after tracking). My SGIE generates feature vector for face detected and then I calculate cosine distance for similarity with existing feature vectors from my dataset. If the detected face feature vector is not similar to any of my dataset’s feature vectors, I want to save the image of that particular face detected.

Is there anyway I can do it? I am new to deepstream and I am not able to find anything related to it.

please refer to sample deepstream-mdx-perception-app for object embedding vector.

I am already able to save the embeddings from SGIE. What I want to do is, If my SGIE vectors is not similar to my dataset vectors I want to save the FACE detected by my PGIE.

There is no update from you for a period, assuming this is not an issue any more. Hence we are closing this topic. If need further support, please open a new one. Thanks.
please refer to sample \opt\nvidia\deepstream\deepstream\sources\apps\sample_apps\deepstream-image-meta-test. nvds_obj_enc_process can save objects to the jpg files when NvDsObjEncUsrArgs’s isFrame is 0.

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.